llvm-project.git/llvm/test/Transforms/LoopVectorize/pointer-select-runtime-checks.ll, branch users/meinersbur/flang_runtime_split-headers Unnamed repository; edit this file 'description' to name the repository. [VPlan] Remove reverse() of predecessors from VPInstruction::generate. 2024-12-17T20:44:32+00:00 Florian Hahn flo@fhahn.com 2024-12-17T20:44:31+00:00 4ad0fdd1631eeae432714c03ede01a10dc00025d This was originally done to reduce the diff for the change. Remove it and update the remaining tests. NFC modulo reordering of incoming values. Clean up after https://github.com/llvm/llvm-project/pull/114292. 
This was originally done to reduce the diff for the change. Remove it
and update the remaining tests. NFC modulo reordering of incoming
values.

Clean up after https://github.com/llvm/llvm-project/pull/114292.
[LoopUtils] Freeze compare results for diff checks instead of pointers. 2023-11-21T10:54:36+00:00 Florian Hahn flo@fhahn.com 2023-11-21T10:54:35+00:00 ead35564c0136fef0c1efface7b9857ad87af9de THe freezes are introduced to avoid branch on undef/poison, if any of the pointers may be poison. The same can be achieved by just freezing the compare, which reduces the number of freezes needed. See https://alive2.llvm.org/ce/z/NHa_ud Note that the individual compares need to be frozen and it is not sufficient to only freeze the resulting OR: Result OR frozen only (UNSOUND): https://alive2.llvm.org/ce/z/YzFHQY Individual conds frozen (SOUND): https://alive2.llvm.org/ce/z/5L6Z3f 
THe freezes are introduced to avoid branch on undef/poison, if any of
the pointers may be poison. The same can be achieved by just freezing
the compare, which reduces the number of freezes needed. See
https://alive2.llvm.org/ce/z/NHa_ud

Note that the individual compares need to be frozen and it is not
sufficient to only freeze the resulting OR:

Result OR frozen only (UNSOUND): https://alive2.llvm.org/ce/z/YzFHQY
Individual conds frozen (SOUND): https://alive2.llvm.org/ce/z/5L6Z3f
[LV] Disable runtime unrolling for vectorized loops. 2023-01-06T10:56:17+00:00 Florian Hahn flo@fhahn.com 2023-01-06T10:56:17+00:00 68469a80cb7468c577701d725a56a86454b8f0cb This patch adds metadata to disable runtime unrolling to the vectorized loop. If runtime unrolling/interleaving is considered profitable, LV will interleave the loop directly. There should be no need to perform runtime unrolling at a later stage. Note that we already add metadata to disable runtime unrolling to the scalar loop after vectorization. The additional unrolling unnecessarily increases code size and compile time. In addition to that we have several bug reports of unncessary runtime unrolling for vectorized loops, e.g. PR40961 Compile-time improvements: NewPM-O3: -1.04% NewPM-ReleaseThinLTO: -0.59% NewPM-ReleaseLTO-g: -0.97% https://llvm-compile-time-tracker.com/compare.php?from=ce1be13a868d0f8afa367975558c1a6175cce33a&to=78bc2e67f22e9e10e61cdb6cdac4bb857d95eb1b&stat=instructions:u Fixes #40306. Reviewed By: lebedev.ri, nikic Differential Revision: https://reviews.llvm.org/D115261 
This patch adds metadata to disable runtime unrolling to the vectorized
loop. If runtime unrolling/interleaving is considered profitable, LV
will interleave the loop directly. There should be no need to perform
runtime unrolling at a later stage.

Note that we already add metadata to disable runtime unrolling to the
scalar loop after vectorization.

The additional unrolling unnecessarily increases code size and compile
time. In addition to that we have several bug reports of unncessary
runtime unrolling for vectorized loops, e.g. PR40961

Compile-time improvements:

  NewPM-O3: -1.04%
  NewPM-ReleaseThinLTO: -0.59%
  NewPM-ReleaseLTO-g: -0.97%

https://llvm-compile-time-tracker.com/compare.php?from=ce1be13a868d0f8afa367975558c1a6175cce33a&to=78bc2e67f22e9e10e61cdb6cdac4bb857d95eb1b&stat=instructions:u

Fixes #40306.

Reviewed By: lebedev.ri, nikic

Differential Revision: https://reviews.llvm.org/D115261
[VPlan] Only generate single instr for loads uniform across all parts. 2022-09-08T13:27:58+00:00 Florian Hahn flo@fhahn.com 2022-09-08T13:27:58+00:00 422cf99161ed5c34a32ce9ad549b5da29ad7906f VPReplicateRecipe::isUniform actually means uniform-per-parts, hence a scalar instruction is generated per-part. This is a potential alternative D132892. For now the current patch only catches cases where the address is trivially invariant (defined outside VPlan), while D132892 catches any address that is considered invariant by SCEV AFAICT. It should be possible to hoist fully invariant recipes feeding loads out of the vector loop region as well, but in practice LICM should do that already. This version of the patch artificially limits this to loads to make it easier to compare, but this restriction should be easily liftable. Reviewed By: reames Differential Revision: https://reviews.llvm.org/D133019 
VPReplicateRecipe::isUniform actually means uniform-per-parts, hence a
scalar instruction is generated per-part.

This is a potential alternative D132892. For now the current patch only
catches cases where the address is trivially invariant (defined outside
VPlan), while D132892 catches any address that is considered invariant
by SCEV AFAICT.

It should be possible to hoist fully invariant recipes feeding loads out
of the vector loop region as well, but in practice LICM should do that
already.

This version of the patch artificially limits this to loads to make it
easier to compare, but this restriction should be easily liftable.

Reviewed By: reames

Differential Revision: https://reviews.llvm.org/D133019
Recommit "[LAA] Initial support for runtime checks with pointer selects." 2022-06-17T19:06:26+00:00 Florian Hahn flo@fhahn.com 2022-06-17T19:06:25+00:00 e9cced27390ba38eac1144aa1240281a1edadec0 This reverts commit 7aa8a678826dea86ff3e6c7df9d2a8a6ef868f5d. This version includes fixes to address issues uncovered after the commit landed and discussed at D11448. Those include: * Limit select-traversal to selects inside the loop. * Freeze pointers resulting from looking through selects to avoid branch-on-poison. 
This reverts commit 7aa8a678826dea86ff3e6c7df9d2a8a6ef868f5d.

This version includes fixes to address issues uncovered after
the commit landed and discussed at D11448.

Those include:

* Limit select-traversal to selects inside the loop.
* Freeze pointers resulting from looking through selects to avoid
  branch-on-poison.
[LV] Add additional tests for pointer select support. 2022-06-01T20:19:03+00:00 Florian Hahn flo@fhahn.com 2022-06-01T20:19:03+00:00 72aca94b9080f1c89b2772b3a2612022525bf0d0 Additional test cases for D114487. 
Additional test cases for D114487.